Current Server Time: 17:42 (Central Europe)

#apertus IRC Channel Logs

2017/01/06

Timezone: UTC


01:13
hozer
left the channel
01:13
philippej
left the channel
01:13
danieel
left the channel
01:14
_florent_
left the channel
01:14
rexbron_
left the channel
01:14
skinkie
left the channel
01:14
philippej|away
joined the channel
01:14
skinkie_
joined the channel
01:14
philippej|away
changed nick to: philippej
01:14
rexbron
joined the channel
01:14
danieel
joined the channel
01:25
_florent_
joined the channel
02:10
Bertl_oO
off to bed now ... have a good one everyone!
02:10
Bertl_oO
changed nick to: Bertl_zZ
02:21
dimaursu16
left the channel
02:24
dimaursu16
joined the channel
02:24
dimaursu16
left the channel
02:24
dimaursu16
joined the channel
03:04
Spirit532
joined the channel
03:55
dimaursu16
left the channel
06:10
skinkie_
left the channel
06:10
skinkie
joined the channel
06:16
dimaursu16
joined the channel
06:16
dimaursu16
left the channel
06:16
dimaursu16
joined the channel
08:23
slikdigit
left the channel
08:23
slikdigit
joined the channel
08:25
Spirit532
left the channel
08:25
Spirit532
joined the channel
08:31
se6astian|away
left the channel
08:31
intracube_afk
left the channel
08:31
illwieckz
left the channel
08:31
Bertl_zZ
left the channel
08:31
morrigan_
left the channel
08:31
Maqs
left the channel
08:31
fredy
left the channel
08:31
comradekingu
left the channel
08:37
intracube_afk
joined the channel
08:37
illwieckz
joined the channel
08:37
Maqs
joined the channel
08:37
Bertl_zZ
joined the channel
08:37
morrigan_
joined the channel
08:37
fredy
joined the channel
08:37
se6astian|away
joined the channel
08:37
comradekingu
joined the channel
08:40
mithro
left the channel
08:54
mithro
joined the channel
08:54
dimaursu16
left the channel
09:05
Bertl_zZ
changed nick to: Bertl
09:05
Bertl
morning folks!
10:21
se6astian|away
changed nick to: se6astian
10:51
Bertl
changed nick to: Bertl_oO
13:37
se6astian
great news everyone: i figured out the openpnp bottom vision axis orientation issue and can now place 0402 with near 100% perfect accuracy
13:37
se6astian
repeatedly
14:02
se6astian
there is still a small offset somewhere
14:02
se6astian
but that will be easy to correct
14:02
se6astian
https://cloud.gerade.org/index.php/apps/gallery/s/aDVX7UjLykjJeAX#0402-BV-placements-success.jpg
14:11
comradekingu
yay
14:43
Spirit532
se6astian, did you have to choose the jankiest-looking board for the test? :)
14:43
Spirit532
so much schmoo
14:43
Spirit532
but congrats
14:46
kaiserlich
joined the channel
14:46
se6astian
the double sided tape and gaveyard of resistors make it indeed look aweful :D
15:07
comradekingu
was hoping that was not the usual run of the mill :)
15:08
Spirit532
editing high speed video from a mildly aged camera is such a pain
15:08
Spirit532
their proprietary format only really usefully exports into uncompressed avi and raw tiffs, usefully
15:08
Spirit532
grrrr
15:08
Spirit532
at least resolve understands tiff sequences and plays them blazing fast
15:09
se6astian
fun!
15:09
Spirit532
I still haven't figured out how to do an average FPN removal
15:09
Spirit532
the sensor is a bit out of cal and has some black patterns
15:10
Spirit532
I know what I need to do(take an average of X black frames, divide by number, subtract), but I have no idea how to do it in any NLE/grading software
15:11
se6astian
we also tried that approach with the beta first
15:11
Spirit532
apparently I got autobanned in ##electronics, hold on
15:11
Spirit532
left the channel
15:11
Spirit532
joined the channel
15:11
se6astian
but did not get to a point where it produced anything useful
15:12
se6astian
alex then approached the topic from a programming POV with proper math background
15:12
se6astian
and that worked out very well
15:12
Spirit532
well, I can go inside the $60 000 camera and tweak the(presumably) 14 column amplifier voltage pots
15:12
Spirit532
but I'd rather not, seeing the numbers
15:13
se6astian
interesting that they actually expose such controls :)
15:13
Spirit532
expose?
15:13
Spirit532
who said anything about expose!?
15:13
Spirit532
I ripped into the camera, I own it!
15:13
se6astian
ha :D
15:13
Spirit532
I wouldn't be an EE if I didn't take it apart.
15:15
Spirit532
which alex is the one that made the FPN magically woosh away?
15:22
se6astian
alexML:
15:31
kaiserlich
left the channel
15:35
alexML
yes?
15:38
Spirit532
alexML, teach me the ways of fixing FPN
15:41
jucar
joined the channel
15:46
alexML
these are my notes for the Beta: https://wiki.apertus.org/index.php/Pattern_Noise
15:46
alexML
are you talking about FPN from the same sensor, or something different?
15:47
Spirit532
something entirely different
15:47
alexML
have some samples?
15:47
Spirit532
FPN from an extremely large pixel(1280x1024, full-frame) sensor from a HS camera
15:47
Spirit532
have samples indeed, hold on
15:48
Spirit532
i.spirit.re/fpnsmp.zip
15:48
Spirit532
there.
15:48
Spirit532
these are raw bitmaps
15:48
alexML
for the beta, a large part of the dynamic row noise (the non-fixed part of the pattern noise) could be estimated from the black columns
15:48
Spirit532
it's not dynamic
15:48
Spirit532
it stays nearly perfectly constant throughout the operation
15:48
Spirit532
just getting slightly higher/lower in intensity with temperature
15:49
Spirit532
it's designed to operate at ~45-48C(silicon, body is at around 50C), but the sensor appears to have been calibrated stone cold
15:49
Spirit532
and so when it gets up to operating temp, it becomes fuzzy.
15:50
intracube_afk
changed nick to: intracube
15:50
alexML
if it's not dynamic, a dark frame subtraction should do the trick
15:50
alexML
it's a bit harder to measure the temperature variations though
15:51
Spirit532
except I can't just subtract that
15:51
Spirit532
at least I wasn't able to in AE/Vegas
15:51
Spirit532
it just goes funny
15:51
alexML
are you working on raw (linear) data?
15:51
Spirit532
I think so
15:52
alexML
what image format do you get from the camera?
15:52
Spirit532
proprietary voodoo, can be exported into tiffs
15:52
Spirit532
https://i.spirit.re/UL0Ln.zip
15:53
Spirit532
frame 0 is black(almost completely), the other is with action, slightly underexposed
15:55
Spirit532
that was shot at ISO 3200-ish, 1/100 000 shutter, f/5.6
15:55
Spirit532
no gain
15:57
pusle
left the channel
15:58
alexML
try this:
15:59
alexML
octave:1> a = imread('match1_1kfps_100kthshut_f56-N001_00000000.tif');
15:59
alexML
octave:2> b = imread('match1_1kfps_100kthshut_f56-N001_00000899.tif');
15:59
alexML
octave:3> imwrite(uint16(b-a), 'd.tif');
15:59
alexML
octave:4> imshow(sqrt(b-a), []);
15:59
Spirit532
implying I have octave D:
15:59
Spirit532
but, that's just subtracting it directly, right?
16:00
Spirit532
and making it non-linear
16:00
alexML
the nonlinear part is just to boost the shadows when previewing
16:00
jucar
left the channel
16:01
Spirit532
the other problem is
16:01
Spirit532
sometimes the shots can't be purely dark
16:01
Spirit532
and I can't do a reference frame every shot
16:01
alexML
it changes over time?
16:01
Spirit532
no, I can't have a black shot for every frame, since I don't have a shutter capable of flipping at 168kHz
16:01
Spirit532
lol
16:02
Spirit532
the pattern is static, but it gets more/less pronounced
16:03
alexML
so it's not static :P
16:04
Spirit532
can that be defined as not static?
16:04
alexML
it changes during a single video clip?
16:04
Spirit532
no, but it changes clip-to-clip
16:04
alexML
even if the clips have the same settings? (exposure time, iso/gain)
16:04
Spirit532
if they're shot almost sequentially, not really
16:05
Spirit532
but if I give the camera an hour to warm up, they vary wildly
16:05
alexML
ok, so it's probably temperature
16:05
Spirit532
yes
16:05
Spirit532
I can actually get the raw sensor temp with each clip
16:05
Spirit532
but I doubt that really helps
16:06
Spirit532
in that clip the sensor was at 44C
16:06
alexML
you could record a series of dark clips (along with temperature metadata), ranging from cold (winter helps) to hot (say after 1-2 hours of recording)
16:07
Spirit532
that's kinda what I did with the first archive
16:07
Spirit532
but it's never spot-on
16:07
alexML
you may find hints about temperature by measuring hot pixels
16:08
alexML
https://wiki.apertus.org/index.php/Black_Calibration#Dark_current_measurement_from_hot_pixels
16:08
Spirit532
if you scroll through the images blown way up in gain
16:08
Spirit532
https://i.spirit.re/OGZx7.webm
16:08
Spirit532
you can see that there's regular CMOS sensor noise, and then there's this "filter" on top
16:08
alexML
although, this alone was not enough on the Beta to fix temperature variations
16:08
Spirit532
which remains absolutely constant
16:09
Spirit532
I guess my best bet is to just take a calibration shot before every real shot as close as possible
16:10
alexML
looks like some flicker; I'm not sure what I'm looking at (e.g. what's on the time axis, how the grayscale mapping was done)
16:10
Spirit532
that's just flicking through the images in irfanview
16:10
Spirit532
it does a min-max thing
16:10
Spirit532
to show you the most image
16:11
alexML
ah, so that is very likely to cause flicker
16:11
Spirit532
there's no real flicker
16:12
toxitobi
joined the channel
16:17
Spirit532
yes
16:18
alexML
afk
16:33
kaiserlich
joined the channel
16:33
Spirit532
left the channel
16:52
kaiserlich
left the channel
17:15
kaiserlich
joined the channel
17:22
slikdigit_
joined the channel
17:31
kaiserlich
left the channel
18:06
kaiserlich
joined the channel
18:09
pusle
joined the channel
18:34
kaiserlich
left the channel
18:51
kaiserlich
joined the channel
19:21
morrigan1
joined the channel
19:21
Bertl_oO_
joined the channel
19:22
illwieckz_
joined the channel
19:22
intracube
left the channel
19:22
illwieckz
left the channel
19:22
Bertl_oO
left the channel
19:22
morrigan_
left the channel
19:44
kaiserlich
left the channel
20:00
dimaursu16
joined the channel
20:00
dimaursu16
left the channel
20:00
dimaursu16
joined the channel
20:00
dimaursu16
left the channel
20:06
kaiserlich
joined the channel
20:08
intracube
joined the channel
20:38
kaiserlich
left the channel
20:59
kaiserlich
joined the channel
21:02
slikdigit
left the channel
21:02
slikdigit_
changed nick to: slikdigit
21:03
slikdigit_
joined the channel
21:31
kaiserlich
left the channel
21:38
dimaursu16
joined the channel
21:38
dimaursu16
left the channel
21:38
dimaursu16
joined the channel
22:08
dimaursu16
left the channel
22:24
arpu
left the channel
22:26
se6astian
changed nick to: se6astian|away
22:37
arpu
joined the channel
22:42
toxitobi
left the channel
23:08
slikdigit__
joined the channel
23:10
slikdigit
left the channel
23:10
slikdigit__
changed nick to: slikdigit